Party-Line Voting and Committee Assignments
in the German Mixed Member System
Thomas Stratmann
Department of Economics
George Mason University
[email protected]
Traditionally, electoral systems fall into two categories: proportional and majoritarian
systems. Proportional systems tend to have multi-seat districts, party lists, and multiple parties,
while majoritarian systems tend to have geographically based single-seat districts, district
representatives who are chosen by plurality rule, and a two-party system. A relatively new
electoral system is the mixed-member electoral system where one fraction of the legislators are
elected through a majoritarian procedure, and where the other fraction through a proportional
procedure.
The consequences of electoral systems for political and economic outcomes have been a
long-standing research topic in political science, and recently economists have started examining
their economic effects (Persson, Roland and Tabellini 2000). In contrast to electoral systems that
select legislators solely through one electoral rule, mixed-member electoral systems allow
scholars to answer long-standing questions regarding the importance of electoral rules for
legislative behavior, and subsequently, for political and economic outcomes, without
confounding country effects with electoral rule effects.
An ideal analysis of electoral rules on legislative behavior would randomly assign
candidates to one or another electoral rule under which they would be elected. Such random
assignment allows one to determine a causal effect of rules on behavior. Without random
assignment, and when estimating the effects of electoral rules across countries, there is the
possibility that unobserved country characteristics that are important for legislator behavior are
correlated with the electoral rule in the country. If countries use solely a proportional electoral
system or solely a majoritarian system it is impossible to disentangle the effect of country
specifics from the effect of the electoral rule on legislative behavior, since unobserved country
1
characteristics may be correlated with the electoral rule adopted. Some of these issues can be
avoided when one examines countries with mixed member systems, i.e. countries in which some
legislators are elected through one electoral method and others through a different method.
Only a few countries have a mixed member system, although there has been a tendency
for countries to adopt a mixed system within the last fifteen years. For example Israel, Italy,
Japan, New Zealand and Venezuela have moved from having all of their members elected
through solely one electoral system to a mixed member system (Shugart and Wattenberg 2001).
One mixed member system is in Germany, and this paper examines the behavior of German
legislators in the 1990s.
Most of the literature comparing electoral rules has looked at the macro picture, such as
the differences between proportional representation (PR) and a majoritarian system, also known
as first-pass-the-post (FPTP), with respect to the number of parties, party strength, their
differences with respect to the viability of small parties, representation of preferences, and
legislative majorities (see, for example, Rae 1971).1 Few studies, however, have analyzed the
micro level, namely the incentives faced by legislators elected under one rule versus another.
Examining differences in legislative behavior by electoral rule is of interest because legislative
behavior is one of the central inputs into political and economic outcomes.
One of the key responsibilities of legislators is to cast votes on a variety of bills and
amendments. Relative to the majoritarian electoral systems where parties are relatively weak,
there is a high level of party unity voting in proportional systems. Also in mixed member
1
Virtually all of these studies are subject to the simultaneous equation bias mentioned
earlier.
2
systems as for example in the German system, legislators tend to vote with their parties and most
votes are voice votes. However, documented votes which record who each legislator voted exist
in the German system. This paper test the hypothesis that legislators who are elected through the
majoritarian system are more likely to deviate from the party line than legislators elected through
a proportional system.
To my knowledge only one previous study has done a quantitative examination of roll
call votes in Germany. This study derives a measure of party discipline from roll call votes over
the period from 1949 to 1990 and relates this measure to the homogeneity of party members
socio-economic characteristics (Saalfeld 1995). This study did not find systematic patterns in the
data, and Saalfeld did not examine voting behavior by the type of electoral rule through which
each German legislator became a member of parliament.
Besides voting, legislators craft bills, and this occurs in committees of the German
Bundestag. Committees allow legislators to influence language in bills to the benefit of
particular groups in society. Legislator have incentive to support different groups, depending on
the electoral system through which they were elected. For example, legislators elected though
the FPTP electoral rule are predicted to seek benefits for their home constituency, also referred to
as pork-barreling (Weingast, Shepsle and Johnsen 1981). While this pork barrel activity has
been documented in the United States, it is not clear whether it is primarily attributable to the
electoral rule FPTP, used for electing U.S. legislators, or to other characteristics of American
politics.2 This paper tests whether German FPTP legislators seek committees that put then in an
2
Within the context of the United States, politicians’ incentives to secure federal funds for
their home districts has been modeled, for example, by Weingast 1979; Shepsle and Weingast
(1981); Weingast, Shepsle, and Johnsen (1981); Ferejohn and Krehbiel (1987); and Inman and
3
advantageous position to deliver central government funds to their home constituency. This
paper also tests whether PR legislators, whose reelection depends on their ranking in the party
and whether they are supported by groups supporting the party will seek committees that put
them in a position to redistribute resources to supporters of their party.
Previous work has found support for the hypothesis that FPTP legislators are members of
committees that are particularly suitable to help their home constituency and that PR legislators
are members of committees that help their party consistency (Stratmann and Baur 2002). This
paper tests this hypothesis for the one electoral period for which I analyze party unity voting.
I find that FPTP legislators are more likely to vote against the party than their PR
counterparts. Further, FPTP legislators are predominantly on committees where they have
influence over the allocation of benefits to their geographic reelection constituency, while PR
legislators are members of committees that have control over funds that benefit party reelection
constituencies. Thus, electoral rules have a significant influence on day-to-day legislative
behavior, which in turn effects government policy.
The next section of this paper discusses the underlying institutions in Germany. The third
section presents hypotheses, and data are described in the fourth section. Results are presented in
section five and section six concludes.
II. Institutional Background
A. Electoral System
Fitts (1990). Evidence for the effectiveness of pork barrel spending for securing reelection has
been documented by Levitt and Snyder (1997).
4
In Germany’s electoral system some representatives are elected through plurality rule
from single member districts and others though proportional representation. Each voter casts a
candidate ballot (Erststimme) for the district candidate and party-list ballot (Zweitstimme) for the
party.3 For the party list ballot the vote does not vote for the party, not a particular member on
the party-list. The district candidate with the largest number of votes wins and becomes a
member of parliament. Thus, the fact that the German electoral system is party majoritarian,
stems from the fact that the winner of the election in each of the n single member districts
becomes a member of parliament.
Each party receives that fraction of the seats in parliament that is proportional to the
percent of votes that the party receives on the party-list ballot. Those seats are allocated first to
the winners of district races. If the party receives more seats through the party-list ballot than it
can fill with winners in district races, party members with the highest rank on the state party list
fill those slots. Thus, the number of district races won does not tend to influence the overall
number of seats obtained by the party, except when a party wins more races than the number of
seats allocated through the party list ballot. In the latter case, the party can keep the extra seats
(Überangsmandate). District candidates may also appear on the party-list ballot. If those
candidates win in the district election, they enter the parliament based on winning the district
election, not based on having a slot on the party list.
The party ballots are totaled at the state level. Suppose the state of Baden-Württemberg
contributes 100 seats to the parliament, which are divided into 50 district seats (implying that this
state has 50 district races) and 50 party-list seats. Further, suppose the CDU wins 45 percent of
3
For an analysis of voters’ responses to this two-vote system see Bawn (1999).
5
the party-list ballot votes, the SPD win 42 percent, and the FDP and Greens receive 7 and 6
percent, respectively. Lastly, assume that the CDU wins in 40 of the 50 district races while the
SPD wins in the remaining 10 races. In this case the CDU’s district winners fill 40 of the 45
seats and 5 individuals are chosen (those with the highest rank) from the CDU’s state party list.
Similarly, the SPD district winners fill 10 of the 42 seats allocated to the SPD, and the remaining
seats are filled from the top spots of the SPD’s state party list. Thus, the larger the number of
FPTP seats won by a party, the fewer the PR seats it receives; therefore the mix of FPTP and PR
seats differs for each party. In this example, the FDP and Greens draw their 7 and 6
representatives entirely from their state lists.
If a party has more FPTP seats than it has representatives through the PR system, it can
keep these extra seats (Überhangsmandate). In the previous example, if the CDU receives 47
seats through the candidate ballot, it can keep the extra two seats and no individuals are chosen
from the state party list.4 For the purpose of the subsequent empirical analysis of committee
assignments, this circumstance is of little or no quantitative importance.5
In summary, the legislature has n + m seats, where m is the number of legislators chosen
from state party lists, and n is the number of plurality rule winners in the district races. Since n is
roughly equal to m, approximately half of the legislators are elected by plurality rule (n
legislators) and the other half is elected by proportional representation from state (Länder) party-
4
In this case, the overall number of seats in the legislature is adjusted.
5
In the 1998, 1994 and 1990 elections the SPD had 13, 4, and 0 extra seats, and the CDU
had 0, 12, and 6 extra seats, respectively. Neglecting those extra seats, the German parliament
has 656 legislators. Prior to the German unification, the West German parliament had 496 seats.
6
lists. Germany is divided into 16 states.6 Each state (s) contributes ks seats to the legislature,
where 3ks = n + m. The number of seats for each state is determined by the state’s population
size. The party-list ballot determines the final distribution of the n+m seats in the legislature.
That is, the mix of FPTP and PR seats differs for each party, depending on the number of district
races won, since each party’s overall number of seats is determined by the party-list ballot.
Legislators who are elected through PR or FPTP have identical rights and obligations within the
legislature.
B. Voting
Voting in the German Bundestag occurs either by secret ballot or by voice vote where
simply the number of yea and nay votes are counted (Geschäftsordnung paragraph 48 and 51,
einfache Abstimmung, Zählen der Stimmen) or the recorded roll call vote (Geschäftsordnung
paragraph 52 - namendliche Abstimmung. The recorded roll call vote occurs when one party
(Fraction) requests a recorded vote or when five percent of the representatives request a recorded
vote. Prior to 1980 a different parliamentary rule applied. That rule required a request from a
minimum of fifty legislators in order to hold a roll all vote.
The most frequent vote type in the German Bundestag is the voice vote which does not
provide information as to who voted in favor or against the proposition. The literature on
recorded votes in the German Bundestag suggests that recorded votes occur when a party expects
to achieve political gain from a recorded vote, or when it wants to stake out a clear position, or
6
Prior to the unification of East and West Germany, West Germany was divided into 11
states.
7
wants to assure that interest groups and voters know about the party’s position (Loewenberg
1971, Borgs-Maiejewski 1986)
C. Committee System7
The German legislature, the Bundestag, has permanent (standing) committees and also
has special committees and investigatory committees. The issue areas of standing committees
mirror, for the most part, the policy issues for which the government ministries are responsible.
Legislation is referred to the committees from these the executive and from the
parliamentary parties. Committees are required to make a recommendation whether or not the
bill should be passed to the entire legislature. This recommendation may include amendments to
the legislation. Bills are modified at the committee stage in about 60 percent of all cases. Often
bills are in the committee for six months (Johnson, p.124, 1979)8 during which time they
committee may hold public hearings during which experts and affected parties provide input. In
addition to modifying legislation, committees provide “infrastructure for communications and
information between members of parliament, government ministers, bureaucrats, and interest
groups” (Saalfeld 1998, p.58).9
“Committees [in the German legislature] play a major part in giving ... legislative output
its final shape” (Johnson, p.135, 1979). As in the United States, committee members’ decisions
7
This section draws on the discussion of German committees in Stratmann and Baur
8
In contrast to the United States’ practice, subcommittees are virtually non-existent.
(2002).
9
In international comparison, German committees are more powerful than British House
of Commons committees, and less powerful than American congressional committees.
8
are typically influenced by party positions. Organized interests attempt to influence legislation
by lobbying committee members. Interests also have influence, in part, because committee
members are often interest group officials. In fact, appointments to committees are often made
so as to match legislators who are members of pressure groups with the jurisdiction of
committees (Saalfeld 1998, p.58). For example, the education committee has been dominated by
members from the academic community (Johnson 1979).
Committees also work on non-legislative items such as conducting investigations,
holding hearings on particular issues of concern, and prepare reports. They do not only work on
issues that are referred to them by the Bundestag, but they also develop their on initiative with
respect to holding hearings on matters that they view as important, and consider matters as they
see fit. During the 12th Bundestag, there have been 3,024 committee meetings (Hübner 1995).
German committees also have some oversight functions with respect to the executive. For
example committees investigate government regulations, review activities of the executive
branch of government (Johnson 1979).
Committees assignments are made after the general election, at the beginning of a new
legislative session. First legislators indicate their preferences regarding committee assignments
and next the party whips formulate a proposal to the party caucus (Fraktion). The party caucus
decides on the final committee assignments (Kaack 1990). Parties allocate seats according to
personal preferences of legislators, representativeness of various political groups within the
party, and representativeness of the various German regions (Johnson 1979, Ismayr 1992, p.189).
If there exists a greater demand for seats on a particular committee than there are slots, intra-
9
party bargaining occurs to achieve an agreement.10 FPTP members appear to be in a better
bargaining position and are more likely to obtain their first choice of committee seat. “Where
they [FPTP members] really want to have it their way, mostly in matters of interest to their
voting districts, they can...mobilize considerable bargaining power within their parliamentary
groups” (Patzelt 2000, p.38).
Committees are assigned after the election, at the beginning of a new legislative session.
Legislators indicate their committee preferences to their party, and the party caucus determines
the final committee assignments (Kaack 1990). Legislators can be fairly independent in choosing
their specialization (Hübner 1995). Parties allocate seats according to personal preferences of
legislators, representativeness of various political groups within the party, and representativeness
of the various German regions (Johnson 1979, Ismayr 1992, p.189). Appointments to
committees are often made so as to match legislators who are members of pressure groups with
the jurisdiction of committees (Saalfeld 1998, p.58). For example, the education committee has
been dominated by members from the academic community (Johnson 1979).
Once each legislator has voiced his preference, party whips formulate a proposal for
committee assignments which requires consent by the party caucus (Fraktion). If there exists a
greater demand for seats on a particular committee than there are slots, intra-party bargaining
occurs to achieve an agreement.11 FPTP members appear to be in a better bargaining position
10
Demand for committee seats may not exceed the supply of seats if legislators do not
request assignments they do not get in any case. If this is the case, the self-selection aspect of
choosing committee seats is overstated.
11
Demand for committee seats may not exceed the supply of seats if legislators do not
request assignments they do not get in any case. If this is the case, the self-selection aspect of
choosing committee seats is overstated.
10
and are more likely to obtain their first choice of committee seat. “Where they [FPTP members]
really want to have it their way, mostly in matters of interest to their voting districts, they
can...mobilize considerable bargaining power within their parliamentary groups” (Patzelt 2000,
p.38).
Similarly to the United States, traditionally, reelected legislators can retain their seats on
the committee on which they previously served. Once a legislator “has secured a committee
assignment to his liking, he can usually count on holding on to it” (Johnson, p.119, 1979). Most
legislators are members of only one committee. While multiple committee membership is more
common among members of smaller parties, a significant fraction of CDU, CSU, and SPD
legislators are members of multiple committees. In the 1990s, approximately fifteen percent of
the legislators from major parties have been members on multiple committees, while over fifteen
percent of the legislators from major parties have had no committee assignments (as Ordentliche
Mitglieder).
The number of seats for each party on each committee is proportional to the share of seats
in the Bundestag, rather than the share of the popular vote. Each committee has one chairperson
who can be from any party. A party’s overall number of committee chairs is determined by its
share of party-list ballot votes. For example, if the Christian Democratic Party receives 40
percent of the popular vote, 40 percent of all committee chairpersons come from that party.
III. Hypotheses
Pervious work on mixed member systems has analyzed the determinates as to why
countries switch from one type of electoral system to a mixed member system (Shugart and
11
Wattenberg 2001). A few studies analyze the consequences of mixed member systems with
respect to economic or political outcomes. Two examples of such recent studies are Benoit
(2001) who suggests that directly elected candidates in Hungary are more likely to respond to
party preferences than to district preferences, and Stratmann (2001) who finds that between 1990
and 2002 German legislators have systematically different committee assignments, depending on
whether they were elected from the district or obtained a seat in parliament though the party list.
In contrast, Weldon (2001), suggests that in Mexico the possible effects of a mixed system on
legislative behavior are overwhelmed by other institutional factors. Other work on mixed
member systems has analyzed the effects of split ticket voting (see, for example, Bawn 1999)
Most of the literature on the German parliamentary system has claimed that there are no
significant behavioral differences between both legislator types in Germany (see Nohlen 1990,
Thaysen, 1990, Ismayr 1992). For example, using survey data, Ismayr (1992) finds that PR
legislators are approximately as likely to be active in their geographic district as are FPTP
legislators, lending support to the hypothesis that there is little behavioral difference between the
legislator types and electoral systems. Little emphasis has been placed on the fact that German
FPTP and PR legislators face different reelection constraints that may manifest themselves in
behavioral differences (Patzelt 2000). One survey study, however, shows that FPTP legislators
perceive it as more important to provide projects for reelection purposes than do PR legislators
(Lancaster and Patterson 1990).12
German PR legislators are chosen from state party-lists made up prior to the election.
12
FPTP members more frequently claimed success in influencing project allocations than
PR legislators (Lancaster and Patterson 1990).
12
State party congresses determine the ranking of legislators on the lists. The higher a party
member’s rank on the list, the higher the probability that she or he will become a member of
parliament. The ranking on the list is determined, in part, by the party members’ prominence,
seniority, interest group approval of the nomination, and evidence of longstanding prior party
activities (Conradt 1986, p.124). The top spots on the list are typically reserved for party
leadership. Much bargaining occurs to achieve a high rank, and even candidates who run in
districts that are considered “safe” seek a high list position, to assure their entry into the
legislature in case they are defeated in their district (Conradt 1986).
Voting behavior
PR legislators’ seat in parliament is thus solely due to the fact that they ranked
sufficiently high on the party list. If they were not to serve the party and interest groups that
support the party, it is likely that they would obtain a lower rank on the party list for the next
election. PR legislators are more dependent on having a good standing in the party as opposed to
FPTP legislators, whose reelection is a function of whether they provide services to their
geographic home constituency. Since PR legislators have a stronger incentive to be loyal to their
party and to follow the party line, I predict that PR legislators are more likely to vote with the
party than FPTP legislators.
Committee Assignments
If so-called pork barrel politics are important, then FPTP legislators will seek committees
that help them to support specific projects in the interest of their geographic reelection
13
constituencies.13 Thus, committees where FPTP legislators can obtain extra money for the home
district are especially attractive to them, and they will avoid committees where they cannot
secure specific benefits for their home constituencies. Servicing their constituency does not
solely imply having seats on pork barrel committees; it also implies seeking committees where an
ombudsman role can be played. These hypotheses imply that the distribution of FPTP and PR
legislators on committees differs systematically across committee types.
As noted while I discussed the voting behavior, PR legislators depend on their state party,
not on voters, for a seat in the legislature. Therefore, PR legislators have incentives to service
the constituency supporting their party. This can be done by being a member of specific
committees. As noted previously, interest groups have an influence on the ranking of candidates
on the party-lists. Given their influence, a PR legislator has an incentive to service these groups
by sitting on a committee that addresses their interests.
The Construction Committee (Ausschuß für Raumordnung, Bauwesen und Städtebau),
Traffic Committee (Ausschuß für Verkehr), and Agriculture Committee (Ausschuß für
Ernährung, Landwirtschaft und Forsten) are perhaps the most clear cut examples of committees
where funds can be channeled to the home district and I classify these committees as district
committees.14, 15 Thus, we predict that FPTP legislators will seek assignments on these
13
In comparing FPTP and PR systems, Rae (1971) hypothesized that “redistributive
effects...are usually most intense under first-past-the-post formulae and least intense under P.R.
formulae” (Rae, p. 103, 1971). A similar hypothesis was developed by Lancaster (1986). Recent
work by Persson, Roland and Tabellini (2000) related electoral systems to government size.
14
Examples of Construction Committee and Traffic Committee work are legislation
regarding the building of public housing and streets, and the allocation of subsidies for public
transportation. Examples of Agriculture Committee work are legislation regarding marketing
issues in German regions; how to implement European Union regulations within Germany; and
14
committees. One caveat applies to the Agriculture Committee, since many of the decisions
regarding agriculture are not made at the national legislative level, but on the European level. I
predict that legislators elected through proportional representation will service their
constituencies on committees that are designed to help distribute funds to their electoral
supporters.
I classify the ‘Women and Youth Committee’ (Ausschuß für Frauen und Jugend) and the
‘Family and Elderly Committee’ (Ausschuß für Familie und Senioren) as party committees, because
they have most clearly group-specific redistributive functions. 16 Legislators elected because they
ranked relatively high on the party-list have an incentive to work for the party’s reelection
constituency. Thus, I predict that PR legislators will be committee members on the Family
Committee and its predecessors.
Further, I classify the Health Committee as a party committee. In Germany, the Health
Committee (Ausschuß für Gesundheit) helps determine national health care prices, serving primarily
as an advisory body to the executive government, with respect to national health care laws. This
committee provides FPTP legislators with few opportunities to channel geographically concentrated
benefits to their districts. Thus, we predict that FPTP legislators will avoid this committee. By
contrast, since the elderly population comprises a large part of the CDU’s reelection constituency,
we predict that PR legislators from the CDU will be committee members on the Health Committee.
whether to restrict some agricultural imports.
15
These committees also perform administrative functions, and are also concerned with
“non-pork” activities.
16
These committees work on legislation regarding women’s rights in the work place, and
gender discrimination issues.
15
I classify the remaining committees as “neutral,” because they allow for only limited
possibilities in terms of either services in the district or party constituency or because they are low
prestige committees with limited redistributive functions. Those committees comprise the Budget,
Finance, Interior, Foreign Affairs, Economics, Environment, Sports, European Union, Election,
Labor, Law, Technology Committees, the Petition Committee, and the Education and Science
Committee.
IV. Data and Estimation issues
I study the allocation of seats on committees in the 1990-94 legislative periods. Five
parties have were elected to the German parliament in the 1990-94 election cycle: the CDU
(Christian Democratic Party outside of Bavaria), the CSU (Christian Democratic Party in
Bavaria), the SPD (Social Democratic Party), the Grüne (Green Party), the PDS (Socialist Party),
and the FDP (Liberal Party). The Grüne and the FDP legislators have only PR seats. Legislators
from the CDU, CSU, SPD, and PDS have obtained their seats either through the FPTP or the PR
electoral system. To test for differences in behavior between representatives elected through
FPTP and PR, the focus can thus only be on these four parties. Of these parties, we will not
examine the PDS, because there is little variation with respect to the FPTP and PR distinction.
Only 1 of the 20 PDS members in 1990 were elected through FPTP.
In the data examined, 18 percent of the members in the German legislature won the
election in their district but were not placed on the party-list. Seven percent of the members did
not run for election in a district but became a member of parliament because they were ranked
above the cut-off number on the state party-lists. As explained below, these legislators are not
part of the empirical analysis due to data limitations. The remaining members ran in a district for
16
election and had a place on the state party-list. Fifty-three percent of those legislators lost the
district election and obtained a seat through the PR electoral rule. The fact that losers in district
elections often obtain a seat in parliament via the state party-list weakens FPTP legislators’
incentives to provide direct geographic constituency service because they have a “safety-net” in
case they lose the election.17 This circumstance puts the odds against finding a significant
difference in the committee seat allocation between both legislator types.
In Germany, the government is formed from members of parliament, but these legislators
(Minister and Staatssekretäre) are not members of any committees in the legislature. Thus, we
exclude legislators who are part of the executive branch from the statistical analysis of committee
membership.
An analysis of electoral systems on legislators’ behavior is subject to at least two
endogeneity concerns. First, unobserved country heterogeneity may be a determinant of the
electoral system as well as economic outcomes. Secondly, legislators who, for example, find the
proportional representation electoral system particularly suitable to their campaigning style, may
decide to run for election on the party list, while they would have chosen not to run for election if
they would have had to run in a district. This selection effect may lead to biased estimates of
electoral rules on outcomes. An analysis of legislator behavior in mixed member systems avoids
the first type of endogeneity, but is still subject to the second type of endogeneity. Thus, mixed
member systems provide an interesting subject of analysis.
The regression estimated is
17
Similarly, PR legislators who run in districts and lose the election sometimes maintain a
district office.
17
Y= f(FPTP, PERCENT IN ELECTION, CONSTITUENCY CHARACTERISTICS, SENIORITY) + ,
Where Y is a measure of legislator behavior, namely deviations from the party line or the
committee assignment. FPTP is an indicator variable defined to equal one when a legislator won
the district election and zero when the legislator entered parliament through the party list. I will
also control for the percent received in the election, as some MPs enter the Bundestag through
the party list, but also run in the local race. Thus these legislators may also have an incentive to
consider their home district in their decisions. I also include seniority of the legislator in the
regression equations. Seniority is the number of legislative periods served in the Bundestag. The
seniority variable equals zero for newcomers (freshmen), one for legislators who have been a
member for two periods, two for legislators who have been a member for three periods, and so
forth. Finally I will control for the constituency characteristics, requiring that I drop legislators
who are solely elected through the party list from the regression analysis. The constituency data
employed are the population employed in industry, the service sector, and farming. Table 1
presents descriptive statistics for the variables employed in the regression analysis.
V. Results
Table 2 presents results where the dependent variable in the Poisson regressions is the
number of times a legislator deviated from the party position. A deviation is defined to occur
when a legislator votes against the majority of his party. The first column presents results where
I include in the regression equation an indicator for whether the legislator was elected through
FPTP or PR, the percent obtained in the district election, seniority, and constituency
18
characteristics. The second column adds party indicators and the third column adds regional
state (Länder) indicators.
The results show that the coefficients on the district election indicators are statistically
significant at the six and seven percent level in columns 1 and 2 and statistically significant at the
five percent when regional indicators are included, as in column 3. The coefficient on the
indicator variable suggests that legislators who won the district elections, are approximately
twenty percent more likely to vote against the party line than legislators who do lost the district
election.
The coefficients on industry and service employment are statistically insignificant in all
specifications and also seniority does not appear to be a determinant of party line voting.
Legislators representing farm constituency, however, are more likely to deviate from the party
line.
Table 3 presents results from the probit regressions that have membership on a district
committee as the dependent variable. The point estimates on the FPTP indicator are statistically
significant at the five percent level, indicating that FPTP legislators are more likely to be
members of committees that allow them to direct funds to their home districts than PR
legislators. Regardless of the specification, the point estimates are similar, suggesting that FPTP
legislators have an almost ten percent higher likelihood to be a member of FPTP committees than
their PR counterparts. The results are robust when the percent of the popular election are
excluded from the regressions.
Of the control variables, none but seniority is statistically significant. One year of extra
seniority reduces the likelihood to belong to a FPTP committee by one to two percent. When the
19
FPTP indicator is included, the coefficient on the percent of the popular election is positive and
statistically significant, indicating that legislators who are elected by large margins are more
likely to be a member of FPTP committees. This is perhaps not too surprising, given that the
FPTP and the percent in the election are positively correlated: the reason a legislator is a FPTP
legislator is that this legislator won the local election.
Table 4 presents the results for determinants of committee membership on the party
interest group committees. Here the coefficient on the FPTP indicator is negative, although the
point estimates are only statistically significant at the ten percent level in column 1 and at the
twelve percent level in column 2. Given that the percent in the election and the FPTP indicator
are positively correlated, I reestimateed the regression excluding the percent in the election.
Those results (not reported) show that the point estimates on the FPTP dummy are negative and
statistically significant at the five percent level, in all specification, and that the point estimates
are of similar magnitude as reported in Table 4. Among the control variables and with exception
of the point estimates on seniority, the point estimates are statistically insignificant.
The results in Table 3, and with some caveats the results in Table 4, are also consistent
with the hypothesis that FPTP and PR legislators seek out different committees in their goal to
increase their chances of reelection. However, the results are also consistent with the hypothesis
that a party assigns committee assignments in a way documented. If parties seek to maximize
seats in the electorate, they would want to put FPTP legislators on committees that would help
them in their reelection campaigns. The committee data underlying the regressions in Tables 3
and 4 do not serve well in discriminating between the“legislator- choice” model and the “partychoice” model, which arrive at similar predictions regarding committee membership.
20
VI. Conclusions
This paper examines legislator behavior, using the variation that is offered by Germany’s
mixed member system. The fact that approximately half of the German legislators are elected
through a majoritarian system and the other half through a proportional system, lends itself to testing
the hypothesis that electoral rules influence legislator behavior. I analyzed legislative behavior with
respect to their voting records and with respect to their committee assignment. With respect to the
voting behavior, I examined how many times legislators are voting against their party. With respect
to committee assignments I examined whether committee assignments differ depending on the
electoral rule under which a legislator entered the German Bundestag. This paper documents that
legislators elected through a majoritarian system (FPTP legislators) are more likely to deviate in their
voting behavior from the party line than legislators elected through the proportional system, and thus
through their ranking on the party list (PR legislators). Also, this paper shows that committee
membership significantly differs between both legislator types. While FPTP legislators are members
of committees that allow them to direct funds to their home district, PR legislators tend to be
members of committees that help them serve their party constituency. For future research, it would
be interesting to study whether the same differences can be documented in other mixed member
electoral systems.
21
References
Bawn, Kathleen, “Voter Responses to Electoral Complexity: Ticket Splitting, Rational Voters and
Representation in the Federal Republic of Germany,” British Journal of Political Science,
29, 1999, pp.487-505.
Benoit, Kenneth, “Evaluating Hungary’s Mixed-Member Electoral System,” in: Shugart, Matthew S.
And Martin P. Wattenberg, Mixed-Member Electoral Systemts. The Best of Both Worlds?,
Oxford University Press, 2001.
Blais, André and Louis Massicotte, “Electoral Systems,” in Comparing Democracies, Lawrence
LeDuc, Richard G. Niemi, and Pippa Norris, eds., Sage, Thousand Oaks, CA, 1996.
Borgs-Maiejewski, Herman Parlamentsorganisation: Institutionen des Bundestages und ihre
Aufgaben. Heidelberg: R. v. Decker & C. F. Müller 1986.
Conradt, David P., The German Polity, Longman, New York, 1986.
Ferejohn, John and Keith Krehbiel, “The Budget Process and the Size of the Budget,” American
Journal of Political Science, 28, 1987, pp.297-320.
Grofman, Bernhard “SNTV: An Inventory of Theoretically Derived Propositions and a Brief Review
of Evidence from Japan, Korea, Taiwan, and Alabama,” in Elections in Japan, Korea, and
Taiwan under the Single Non-Transferable Vote, Bernhard Grofman, Sung-Chull Lee, Edwin
A. Winckler, and Brian Woodall, eds., Ann Arbor, University of Michigan Press, 1999.
Grofman, Bernhard, Evald Mikkel, and Rein Taagepera, “Electoral Systems Change in Estonia,”
Journal of Baltic Studies, vol. 30, 3, Fall 1999, pp. 227-249.
Hübner, Emil “Parlamentarische Demokratie II,” Heft Nr. 227, Bundeszentrale für politische
Bildung, Germany, 1995.
Inman Robert P. and Michael A. Fitts, “Political Institutions and Fiscal Policy: Evidence from the
U.S. Historical Record,” Journal of Law, Economics, and Organization, 6, 1990, pp.70-132.
Ismayr, Wolfgang, Der Deutsche Bundestag: Funktionen Willensbildung, Reformansätze. Leske &
Budrich, Opladen 1992.
Johnson, Nevil, “Committees in the West German Bundestag,” in Committees in Legislatures: A
Comparative Analysis, Lees, John A. and Malcom Show, eds., Duke University Press,
Durham, NC, 1979.
Kaack, Heino, “The Social Composition of the Bundestag,” in The U.S. Congress and the German
Bundestag, Thaysen, Uwe, roger H. Davidson, and Robert Gerald Livingston, eds, Westview
Press, 1990.
22
Lancaster, Thomas D. “Electoral Structures and Pork Barrel Politics,” International Political
Science Review, January 1986, pp. 67-81.
Lancaster, Thomas D. “Candidate Characteristics and Electoral Performance: A Long-Term Analysis
of the German Bundestag,” in Stability and Change in German Elections, Anderson,
Christopher J. and Carsten Zelle, eds., Westport, Connecticut, 1998.
Lancaster, Thomas D. and W. David Patterson. “Comparative Pork Barrel Politics: Perceptions from
the West German Bundestag,” Comparative Political Studies, January 1990, pp. 458-477.
Loewenberg, Gerhard Parlamentarismus im politischen System der Bundesrepublik Deutschland,
Tübingen: Rainer Wunderlich 1971.
Nohlen, Dieter, Wahlrecht und Parteiensystem: Über die politischen Auswirkungen von
Wahlsystemen. Leske Verlag Gmbh, Opladen 1990.
Patzelt, Werner J., “What can an Individual MP do in German Parliamentary Politics,” in Journal
of Legislative Studies, Vol 5(3), 1999, reprinted in The Uneasy Relationship between
Parliamentary Members and Leaders, Longely, Lawrence D. and Reuven Y. Hazan, eds.,
Frank Cass, London and Portand, OR, 2000.
Persson, Torsten; Roland, G. and Tabellini, Guido, "Comparative Politics and Public Finance,"
Journal of Political Economy, 2000, 108, 1121-1161.
Rae, Douglas W., The political consequences of electoral laws, New Haven, Yale University Press,
1971.
Saalfeld, Thomas, Parteisoldaten und Rebellen, Leske + Budrich, Opladen 1995.
Saalfeld, Thomas, “Germany: Bundestag and Interest Groups in a ‘Party Democracy’, Parliaments
and Pressure Groups in Western Europe, Philip Norton, ed., Cass, Portland, OR, 1998.
Shugart, Matthew S. And Martin P. Wattenberg, Mixed-Member Electoral Systemts. The Best of
Both Worlds?, Oxford University Press, 2001.
Stratmann, Thomas and Marin Baur, “Plurality Rule, Proportional Representation, and the German
Bundestag: How incentives to Pork-barrel differ across Electoral Systems” American Journal
of Political Science, July 2002, pp.506-14.
Thaysen, Uwe, “Representation in the Federal Republic of Germany,” in The U.S. Congress and the
German Bundestag, Thaysen, Uwe, Roger H. Davidson, and Robert Gerald Livingston, eds,
Westview Press, 1990.
Weingast, Barry R., “A Rational Choice Perspective on Congressional Norms,” American Journal of
Political Science, 23, 1979, pp.245-62.
23
Weingast, Barry R., Kenneth Shepsle, and Christopher Johnsen, “The Political Economy of Benefit
sand Costs: A Neoclassical Approach to Distributive Politics.” Journal of Political Economy,
79, pp.642-64.
Weldon Jeffery A. “The Consequences of Mexico’s Mixed-Member Electoral System, 19981997" in Shugart, Matthew S. And Martin P. Wattenberg, Mixed-Member Electoral
Systemts. The Best of Both Worlds?, Oxford University Press, 2001.
24
Table 1
Descriptive Statistics
Means and Standard Deviations
Number of times a MP voted against
his or her party
District Committee
Party Committee
Percent in General Election
Obtained Seat in Parliament by
winning the district election =1, 0
otherwise
Fraction of district working
population employed in industry
Fraction of district working
population employed in farming
Fraction of district working
population employed in services
Seniority
N
(1)
2.237
(1.792)
0.094
(0.293)
0.109
(0.312)
43.953
(8.774)
0.545
(0.499)
0.419
(0.859)
0.033
(0.027)
0.371
(0.072)
2.205
(2.038)
413
Data are for the 1990-1994 German Bundestag.
Table 2
Voting Against the Party Line
Asymptotic Standard Errors in Parentheses
(1)
0.196
(0.106)
(2)
0.195
(0.106)
(3)
0.220
(0.108)
-0.019
(0.006)
0.871
(1.208)
0.036
(0.015)
0.011
(0.015)
-0.017
(0.006)
0.930
(1.211)
0.038
(0.015)
0.011
(0.015)
-0.021
(0.006)
1.256
(1.507)
0.036
(0.018)
0.013
(0.017)
0.021
(0.016)
0.021
(0.016)
0.021
(0.016)
Party Indicators
NO
YES
NO
State (Land) Indicators
NO
NO
YES
N
413
413
413
Obtained Seat in Parliament by
winning the district election =1, 0
otherwise
Percent in General Election
Industry Employment
Farm Employment
Service Employment
Seniority
Log-Likelihood
-772.95
-772.11
-762.01
The dependent variable is the number of times the member of parliament did not vote with the
party. The estimates are derived from Poisson regressions.
Table 3
Membership on “District Committees”
Marginal Effects. Asymptotic Standard Errors in Parentheses
(1)
0.094
(0.042)
(2)
0.097
(0.043)
(3)
0.095
(0.044)
0.001
(0.003)
0.105
(0.460)
0.250
(0.593)
0.092
(0.573)
0.002
(0.003)
0.178
(0.459)
0.400
(0.600)
0.195
(0.574)
0.001
(0.003)
0.437
(0.582)
0.565
(0.728)
0.373
(0.662)
-0.016
(0.007)
-0.016
(0.007)
-0.017
(0.007)
Party Indicators
NO
YES
NO
State (Land) Indicators
NO
NO
YES
N
413
413
413
Obtained Seat in Parliament by
winning the district election =1, 0
otherwise
Percent in General Election
Industry Employment
Farm Employment
Service Employment
Seniority
Log-Likelihood
-118.99
-118.21
-116.87
The dependent variable is whether the member of parliament is on a district committee. The
estimates are derived from Probit regressions.
Table 4
Membership on “Party Committees”
Marginal Effects. Asymptotic Standard Errors in Parentheses
(1)
-0.072
(0.046)
(2)
-0.068
(0.046)
(3)
-0.057
(0.047)
0.001
(0.003)
0.041
(0.482)
-0.373
(0.647)
-0.051
(0.600)
0.000
(0.003)
0.008
(0.480)
-0.358
(0.645)
-0.107
(0.600)
0.000
(0.003)
0.219
(0.581)
-0.603
(0.771)
0.040
(0.668)
-0.034
(0.007)
-0.034
(0.007)
-0.035
(0.007)
Party Indicators
NO
YES
NO
State (Land) Indicators
NO
NO
YES
N
413
413
413
Obtained Seat in Parliament by
winning the district election =1, 0
otherwise
Percent in General Election
Industry Employment
Farm Employment
Service Employment
Seniority
Log-Likelihood
-127.50
-126.59
-125.50
The dependent variable is whether the member of parliament is on a district committee. The
estimates are derived from Probit regressions.